Esta página se muestra en inglés. Una traducción al español está en curso.
SkillsApr 10, 2026·3 min de lectura

Typesense — Fast Typo-Tolerant Search Engine

Typesense is an open-source, typo-tolerant search engine optimized for instant sub-50ms searches. Developer-friendly alternative to Algolia and ElasticSearch.

Listo para agents

Instalación con revisión previa

Este activo requiere revisión. El prompt copiado pide dry-run, muestra escrituras y continúa solo tras confirmación.

Needs Confirmation · 64/100Política: confirmar
Superficie agent
Cualquier agent MCP/CLI
Tipo
Skill
Instalación
Single
Confianza
Confianza: Established
Entrada
step-1.md
Comando con revisión previa
npx -y tokrepo@latest install 92da8a8c-352f-11f1-9bc6-00163e2b0d79 --target codex

Primero dry-run, confirma las escrituras y luego ejecuta este comando.

TL;DR
Typesense delivers sub-50ms typo-tolerant search with faceting, geo search, and vector search out of the box.
§01

What it is

Typesense is an open-source search engine built in C++ that delivers sub-50ms search results with typo tolerance out of the box. It provides faceted filtering, geo search, vector/semantic search, and a JavaScript InstantSearch UI.

Typesense targets developers who need a search backend that is simpler to operate than ElasticSearch and more affordable than Algolia, while still delivering fast, relevant results.

§02

How it saves time or tokens

Typesense requires minimal configuration compared to ElasticSearch. There is no cluster management, shard allocation, or mapping complexity. You create a collection with a schema, index documents, and search. Typo tolerance, ranking, and stemming work without tuning.

§03

How to use

  1. Run Typesense with Docker:
docker run -d --name typesense \
  -p 8108:8108 \
  -v typesense-data:/data \
  typesense/typesense:latest \
  --data-dir /data \
  --api-key=your-api-key \
  --enable-cors
  1. Create a collection and index documents via the REST API.
  1. Search using the API or the InstantSearch.js UI library.
§04

Example

# Create a collection
curl 'http://localhost:8108/collections' \
  -X POST \
  -H 'X-TYPESENSE-API-KEY: your-api-key' \
  -d '{
    "name": "tools",
    "fields": [
      {"name": "title", "type": "string"},
      {"name": "description", "type": "string"},
      {"name": "stars", "type": "int32"}
    ],
    "default_sorting_field": "stars"
  }'

# Search with typo tolerance
curl 'http://localhost:8108/collections/tools/documents/search?q=typsnse&query_by=title,description' \
  -H 'X-TYPESENSE-API-KEY: your-api-key'
§05

Related on TokRepo

§06

Common pitfalls

  • Typesense keeps the entire index in RAM for speed. Ensure your server has enough memory for your dataset size.
  • The API key passed at startup is the admin key. Create scoped search-only keys for client-side use to prevent unauthorized writes.
  • Typesense does not support multi-node clusters in the open-source edition. Typesense Cloud provides managed clustering.

Preguntas frecuentes

How does Typesense compare to Algolia?+

Typesense is open-source and self-hosted, while Algolia is a managed SaaS. Both provide typo-tolerant instant search. Typesense is free to run on your own infrastructure; Algolia charges based on search operations and records.

Does Typesense support vector search?+

Yes. Typesense supports vector search alongside keyword search. You can store vector embeddings in a collection and perform nearest-neighbor searches, enabling semantic/hybrid search use cases.

What languages have Typesense client libraries?+

Typesense provides official client libraries for JavaScript/TypeScript, Python, Ruby, PHP, Java, Go, C#, and Dart. There is also an InstantSearch.js adapter for building search UIs.

Can Typesense handle millions of documents?+

Yes. Typesense can index and search millions of documents with sub-50ms latency. Performance depends on available RAM since the index is kept in memory.

Does Typesense support multi-tenancy?+

Yes. You can use scoped API keys to restrict search to specific collections or filter conditions, enabling multi-tenant search where each tenant only sees their own data.

Referencias (3)

Discusión

Inicia sesión para unirte a la discusión.
Aún no hay comentarios. Sé el primero en compartir tus ideas.

Activos relacionados